Heart Disease Diagnosis Explorer

A clean diagnostic dashboard focused on how core clinical variables distribute across disease status, alongside a ranked view of the strongest model discriminators.

Patient records
Individual observations included in the processed cohort used for all plots and rankings.
Disease prevalence
Overall share of patients labeled with heart disease in the binary diagnosis target.
Sex-specific rate
Diagnosis rate shown as male versus female prevalence, to preview the age-by-sex interaction heatmap.
Strongest discriminator
Highest absolute model coefficient among all ranked predictors in the feature-importance file.
Why these chart types fit this clinical question

This layout follows the visualization policy by keeping every analytical graphic full-width, limiting each view to a manageable number of series, and pairing the encoding to the data structure rather than forcing one generic chart type.

  • Horizontal ranking bars work best for the 13 feature coefficients because long feature names remain readable and the ordered layout highlights the strongest discriminators immediately.
  • Violin plots are used for thalach and oldpeak because they preserve continuous distribution shape while still splitting cleanly by diagnosis.
  • Grouped percentage bars are used for coded categorical variables (cp, ca, thal, exang) because each has only a few levels and side-by-side diagnosis shares are easy to compare.
  • An age-band by sex heatmap is used for disease rate because it compactly shows the interaction of two dimensions without overcrowding the dashboard.
No Disease Disease Heatmap = higher disease rate, darker cell
Ranked feature discriminators
Absolute coefficient magnitude ranks the variables by discriminatory strength. Bar color shows direction: positive values align with disease, negative values align with no disease.
Disease rate across age bands and sex
The heatmap shows diagnosis prevalence for each age-band and sex combination; hover to see the exact rate and sample size behind each cell.
Maximum heart rate achieved (thalach) by diagnosis
Violin plots reveal how the full thalach distribution shifts between patients with and without disease, with box summaries embedded for quick comparison.
ST depression (oldpeak) by diagnosis
This distribution view emphasizes spread, skew, and outlier behavior for oldpeak across the two diagnosis groups.
Chest pain type (cp) distribution by diagnosis
Shares are normalized within each diagnosis group so the pattern of chest-pain categories can be compared directly, independent of group size.
Number of major vessels colored (ca) by diagnosis
Low-cardinality grouped bars keep the vessel-count distribution readable while preserving exact category-level contrast between diagnosis groups.
Thalassemia (thal) and exercise-induced angina (exang) by diagnosis
These two coded categorical variables are paired in one wide subplot layout to stay within the requested chart count while still showing both diagnosis splits clearly.